debian2edam

#!/bin/bash -e


# A routine to facilitate the output to STDERR instead of the default STDIN
function STDERR () {
	cat - 1>&2
}

# echoindent outputs a series of blanks to STDOUT. An optional
# second argument is echoed after those blanks if present.
function echoindent () {
	for i in $(seq 1 $1)
	do
		echo -n " "
	done
	if [ "" != "$2" ]; then
		echo $2
	fi
}

level=0
# helper to properly close an open paranthesis
function closeParenthesis () {
	level=$(($level-1))
	echoindent $level
	echo -n "}"
	if [ -n "$1" ]; then
		echo "# $1"
	else
		echo
	fi
}


function echoTerm(){
        level=$(($level-1))
        echoindent $level
        echo "{\"uri\": \"$1\", \"term\": \"Pippi Langstrumpf\"}"
}

# Key argument indicating the debian directory from which to retrieve all the
# information
pathToDebian=$1
#verbose="yes"
verbose=""

# Variable keeping usage information
USAGE=<<EOUSAGE
debian2edam [--upload] <path to 'debian' directory> 

server=https://
Environment variables:
elixir_cat_username
elixir_cat_password 

EOUSAGE

filename=$(basename "$pathToDebian")
if [ "edam" = "$filename" ]; then
	pathToDebian=$(dirname "$pathToDebian") # upstream
	pathToDebian=$(dirname "$pathToDebian") # debian
fi

if [ -z "$pathToDebian" ]; then
	echo "$USAGE" | STDERR
	echo "E: Please specify debian directory in which to find EDAM annotation." | STDERR
	exit -1
fi

if [ ! -d "$pathToDebian" ]; then
	echo "$USAGE" | STDERR
	echo "E: Could not find directory '$pathToDebian'" | STDERR
	exit -1
fi

if [ ! -r "$pathToDebian/changelog" ]; then
	echo "$USAGE" | STDERR
	echo "E: Could not find a changelog file expected at '$pathToDebian/changelog'" | STDERR
	exit -1
fi

cd $(dirname "$pathToDebian")

edamfile="debian/upstream/edam"
if [ ! -r "$edamfile" ]; then
	echo "$USAGE" | STDERR
	echo "E: Could not access file '$edamfile' from $(pwd)" | STDERR
	exit -1
fi

sourcepackage=$(dpkg-parsechangelog |grep ^Source | sed -e 's/`^Source: //' )
version=$(dpkg-parsechangelog |grep ^Version | cut -f2  -d\  | sed -e 's/-[^-][^-]*//' )

declare -a descriptions
declare -a packages

if [ -n "$debug" ]; then cat debian/control; fi

while read pack; do
	p=$(echo "$pack"|sed -e 's/^[^:]*: *//')
	echo Package: $p
	packages[${#packages[*]}]="$p"
done < <(grep "^Package:" debian/control )

while read desc; do
	d=$(echo "$desc"|sed -e 's/^[^:]*: *//')
	echo Description: $d
	descriptions[${#descriptions[*]}]="$d"
	#descriptions[1]="$d"
	#descriptions="$d"
done < <(grep "^Description:" debian/control )

#echo "DESCRIPTIONS: ${descriptions[*]}"
#echo "PACKAGES: ${packages[*]}"
#echo "DESCRIPTIONS: $descriptions}"
#echo "PACKAGES: $packages"

if [ ${#packages[*]} != ${#descriptions[*]} ]; then
	echo "E: Internal error - expected same number of packages (${#packagesp[*]}) as for their descriptions (${#descriptions[*]})" | STDERR
	exit -1
fi

(
if [ -n "$verbose" ]; then
	for packageno in $(seq 0 ${#descriptions[*]})
	do
		echo "# $packageno"
		echo Packages[$packageno]: ${packages[$packageno]}
		echo Descriptions[$packageno]: ${descriptions[$packageno]}
	done
fi
) | STDERR

prevstate="start";
previndent=0
currentscope=""
currenttopic=""
opentopic=0
openfunction=0
openscope=0
indentlen=0

# Core part of the program
# It reads every line of the EDAM file (see end of loop for the redirection)
# and decides what to print to STDOUT.

while IFS='' read -r line 
do
	if [ -z "$line" ]; then
		echo "Read empty line"
		continue
	fi

	if [ -n "$verbose" ]; then
		echo "line: '$line'" | STDERR
	fi

	# retrieve different parts of the description
	blanks=$(echo "$line"|sed -e 's/^\( *\)\([^ :]\+\): *\([^ ]\+\).*$/\1/')
	type=$(echo   "$line"|sed -e 's/^\( *\)\([^ :]\+\): *\([^ ]\+\).*$/\2/')
	val=$(echo    "$line"|sed -e 's/^\( *\)\([^ :]\+\): *\([^ ]\+\).*$/\3/')

	if echo "$val" | grep -q : ; then
		echo "W: found colon in ID of line '$line' - transscribing to underscore" | STDERR
		val=$(echo "$val"|tr ":" "_")
	fi

	#echo "Indent='$blanks'"
	#echo "Indentlength='$indentlen'"
	#echo "Type='$type'"
	#echo "Val='$val'"

	if [  -n "$currentscope" -a "*" != "$currentscope" -a "summary" != "$currentscope" -a "scope" != "$type" ]; then
		echo "I: Wrong scope ($currentscope) - ignored '$line'" | STDERR
		continue
	fi
	indentlen=${#blanks}

	if [ "scope" = "$type" ]; then
		if [ $openfunction -gt 0 ]; then closeParenthesis "openfunction($openfunction) in scope"; fi
		currentscope="$val"
		resourcename=$sourcepackage
		if [ "*"!=$val -a "summary"!="$val" ];then
			resourcename=$val
		fi

		if [ "summary" != "$val" -a "*" != "$val" ]; then
			echo "I: treatment of multiple scopes not yet implemented" | STDERR
		else
			echo "{"
			# Some decent comparison of package names with scope is not implemented
			level=$((level+1))
			echoindent
			echo "Package $resourcename"
			echoindent
			echo "\"version\": \"$version\","
			echoindent
			echo "\"description\": \"${descriptions[0]}\","
			echoindent
			echo "\"topic\": \"{$currenttopic}\""
			openscope=1
		fi
	elif [ "topic" = "$type" ]; then
		if [ $openfunction -gt 0 ]; then closeParenthesis "openfunction($openfunction) in topic"; openfunction=0; fi
		if [ $openscope -gt 0 ]; then closeParenthesis "openscope($openscope) after loop"; openscope=0; fi
		if [ "start" != "$prevstate" ]; then
			closeParenthesis "topic with prior state - weird"
		fi
		currenttopic="$val"
		# at some laterimplementation point, bits generated here would be cached and then distributed
		# to various lower-level scopes
	elif [ "function" = "$type" ]; then
		if [ $openfunction -gt 0 ]; then
			closeParenthesis "openfunction($openfunction) in function"
			openfunction=0
		fi
		echoindent $level
		echo "{function: [ { \"functionName\":  ["
                echoTerm $val
                echo "] }],"
		level=$((level+1))
		openfunction=1
	elif [ "input" = "$type" -o "output" = "$type" ]; then
                if [ $prevstate = $type ]; then
                        echo "},{"
                fi
                if [ $prevstate = 'function' ]; then
                        echo "\"$type\": [{"
                fi
		echoindent $level
		echo "($type $val)"
	else
		echo "W: unknown type '$type' - ignored" | STDERR
	fi
	prevstate=$type
	#echo "indentlen='$indentlen'"
done < $edamfile 

if [ $openfunction -gt 0 ]; then
	closeParenthesis "openfunction($openfunction) after loop"
	openfunction=0
fi

if [ $openscope -gt 0 ]; then
	#echo "I: treatment of multiple scopes not yet implemented"|STDERR
	closeParenthesis "openscope($openscope) after loop"
	openscope=0
fi

#echo "indentlen='$indentlen'" | STDERR

if [ $opentopic -gt 0 ]; then
	opentopic=0
fi

#for i in $(seq $(($indentlen-$openfunction-$openscope-$opentopic)) -1 1)
#do
#	closeParenthesis "indent $i"
#done